StarWind Software

First let me say, Excellent software you have developed. It saves IT proffesionals lots of money.

I want to take this to the next level. And maybe this can accomplished.

This is the layout. A 2003 cluster printing service. AS you are aware , we need to have a storage and a quorom drive. In order to do "best practice" configuration.. For know, I'm using one Volume for the quorum, spool service and printer drivers.

I have 2 server nodes and 1 server acting as the SAN1 device, usign your Starwind iscsi, and on every node we run the MS iSCSI initiator.

If any of the nodes fail all resources are automaticly carried to the other node.

This is a real achievement. BUT..what happends if the 3rd server, the one configured as the SAN device running the Quorum, fails.. the entire cluster is lost. To the point were I have to redo all the work and create a new cluster and begin from scratch.

So I added another small server SAN2, and configured StarWind to do a RAID 1 mirror. The quorum file know resides on 2 servers in a image file.

No problem.. if one fails, I still have the other image file. But that's the problem. Can Starwind make a the quorum image drive (in this case the image files) failover? meaning, i don't have to recover the image file manually and restart the cluster? Can in fact the other server (SAN2) im using go live in the event my SAN1, dies?

This is crucial..

Thanks

F.Cepeda

StarWind v3.x is unable to do auto-failover of targets to another StarWind service as you requested.

To make your StarWind devices fault-protected you need to built another 2-way MS cluster using a non-iSCSI shared storage.
2 instances of StarWind in active/passive mode will provide HAV storage resources for your client servers/clusters.

Is it possible (w2k8 server), to use two separate iscsi targets for soft raid the Quorum?

Target1.disk1(500MB) -> dynamic disc
Target2.disk1(500MB) -> dynamic disc

make a softmirror with w2k8 server (nativ iscsi initiator supports dynamic discs) as drive Q:

than we have an redundand Quorum.

The only problem is the automatic import with diskpart for the active node.
Maybe it must integrated in a cluster ressource:

No. Such a configuration will never work as RAID-ed node will have no clue where and whom to give control. We'll release special version of our StarPort initiator with RAID functionality embedded. It will have own machanism to work in such a configuration.